Ulhumbrus: On the matter of 5...b5 it is worth quoting Capablanca: <The Ruy Lopez is a difficult defence for Black at the best of times and so deviations from the main line can hardly be recommended>
When Kramnik employed the eccentric looking 6...Bc5 in the game Kasparov vs Kramnik, 2000 Kasparov did not find a way to win. However that does not mean that 6...Bc5 is advisable. Kasparov may have made not the best choice.
Carlsen has employed successfully the move Bg5 more than once and perhaps this is the right course at the right moment. To give one possible example, instead of 13 Na3 going after a miserable little pawn (ie an eventual win) Carlsen's stratagem of 13 Bg5! may be better
If 17...b3 is good this suggests that 16 b3 is advisable
With the move 19 e5 perhaps Topalov wants to clear the fourth rank for his rook on a1 to develop via a4.
The move 19 e5? is however an attacking move in a position where White can hardly claim an advantage. It seems to be Black who has the better of it. The attack is unsound and can be forecast to fail. As Black is ahead in development it is Black who will benefit from the opened lines.
After 20 dxe5 Ashley indicates that White is in trouble: Black has lines opened for his bishops.
21 Ng5? attempts to persist with an unsound attack. This increases Black's advantage to more than a winning degree and could have lost at once.
With 21...g6 Nakamura misses his chance. According to Ashley 21...Bxf2+! wins immediately. Ashley indicates as well that it is still not too late for the move ...Bxf2+ on the next move
After 30 Rxe5 White's QB has occupied the diagonal a1-h8 before Black's KB. Topalov said that Black was lost after 30 Rxe5.
The attack 19 e5 may have been unsound but Black had to find all of the right moves in order to prove it.